We are using dfconv to convert files from EBCDIC to ASCII. After the dfconv conversion, characters inside the ASCII range are stored in 1 byte, but characters outside the ASCII range are stored in 2 bytes, e.g. hex 00 gives 00 00 but 41 gives 41 ('A'). This often happens for COMP fields in our data files. We are using a structure file when doing the conversion.
Why does dfconv use 1 byte for valid ASCII characters and 2 bytes for anything else?

From a command line (either Mainframe Express, Net Express or Studio), use these commands:
      set cobsw=-n
      run dfconv profilename inputfile outputfile

This runtime switch will remain active for the duration of the Command Prompt session you are using. By default, Micro Focus will add an extra x'00' to hexadecimal values lower than x'1F'. By setting the -n switch the extra x'00' is not added to the file.